CHAPTER 6

Physics of Optical Tweezers

Timo A. Nieminen, Gregor Knöner, Norman R. Heckenberg,
and Halina Rubinsztein-Dunlop
Centre for Biophotonics and Laser Science
School of Physical Sciences, The University of Queensland
Brisbane QLD 4072, Australia

I. Introduction to Optical Tweezers

II. History
A. Further Reading
III. Basic Principles
A. Further Reading
IV. Fundamental Theory
V. Transport of Energy, Momentum, and Angular Momentum by Light
A. Further Reading
VI. Computational Modeling of Optical Tweezers
A. Further Reading
References

We outline the basic principles of optical tweezers as well as the fundamental

theory underlying optical tweezers. The optical forces responsible for trapping
result from the transfer of momentum from the trapping beam to the particle and
are explained in terms of the momenta of incoming and reflected or refracted rays.
We also consider the angular momentum flux of the beam in order to understand
and explain optical torques. In order to provide a qualitative picture of the
trapping, we treat the particle as a weak positive lens and the forces on the lens
are shown. However, this representation does not provide quantitative results for
the force. We, therefore, present results of applying exact electromagnetic theory
to optical trapping. First, we consider a tightly focused laser beam. We give results
for trapping of spherical particles and examine the limits of trappability in terms
of type and size of the particles. We also study the eVect of a particle on the beam.
METHODS IN CELL BIOLOGY, VOL. 82 0091-679X/07 $35.00
Copyright 2007, Elsevier Inc. All rights reserved. 207 DOI: 10.1016/S0091-679X(06)82006-6
208 Timo A. Nieminen et al.

This exact solution reproduces the same qualitative eVect as when treating
the particle as a lens where changes in the convergence or divergence and in the
direction of the trapping beam result in restoring forces acting on the particle.
Finally, we review the fundamental theory of optical tweezers.

I. Introduction to Optical Tweezers

Optical tweezers, or the single-beam gradient trap, grew rapidly from a novelty to
a widely used tool in many fields. Apart from allowing the noncontact trapping
and manipulation of microscopic particles including living cells and even organelles
within cells, the ability to make noncontact measurements of forces on the order of
piconewtons has enabled the measurement of mechanical properties of cells, the
kinetics and properties of important biological molecules, and measurements of
forces acting on particles in colloidal suspensions. In this section, we briefly review
the history of optical tweezers, including coverage of new developments in trapping
such as using complex optical fields or microfabricated particles which are starting
to see application as new-generation optical tweezers based tools.
In the following section, we discuss the basic principles of optical tweezers. Next,
we outline the fundamental theory underlying optical tweezers. Finally, we outline
the computational modeling of optical tweezers.1

II. History

One of the first recorded speculations that light might be able to exert mechani-
cal forces on particles was in Kepler’s observations of comet tails (Keplero, 1619).
Interest in this idea continued; for example John Michell, one of the pioneers of
the torsion balance, attempted quantitative measurement of optical radiation
pressure (Hardin, 1966) in the 1700s, and succeeded in showing that sunlight can
be concentrated suYciently to destroy one’s experimental apparatus. Such ideas
entered modern science with Maxwell’s predictions of radiation pressure of light
and electromagnetic waves (Maxwell, 1865, 1873).
However, even after the first experimental observation of radiation pressure by
Ledebew (1902) and Nichols and Hull (1901, 1902, 1903a,b), the topic of electro-
magnetic radiation pressure remained an interesting novelty of physics, rather
than an eVect of practical technological value.
The primary diYculty was that the forces resulting from optical or electromag-
netic radiation pressure are small compared to the power required, and are insuY-
cient to overcome gravitational or frictional forces in most circumstances. Thus,
while it was suggested that radiation pressure might be useful in environments

1
At the end of each section, we direct the interested reader to further reading on the topic.
6. Physics of Optical Tweezers 209

where such resisting forces did not need to be overcome—for example, Friedrich
Zander, a pioneer in the fields of rocketry and spaceflight, suggested the use of
solar sails for space propulsion in the 1920s—such applications would remain
beyond practicality for many decades.
This changed in 1969 when Arthur Ashkin realized that, although the forces
resulting from electromagnetic radiation were small, the forces required to move
small particles were also small (Ashkin, 1997). This prompted experiments demon-
strating the acceleration and trapping of small particles by radiation pressure
(Ashkin, 1970). The earliest traps were not purely optical, as they involved a delicate
balance between optical levitation forces and gravity. This placed a strong restric-
tion on the forces that could be applied—radiation pressure forces greater than the
weight of the levitated particle would result in its ejection from the trap. Meanwhile,
gradient forces were observed in these experiments.
At this point, the history of optical trapping and manipulation of small particles
splits into two branches: the trapping and cooling of atoms, which led to the award
of the 1997 Nobel prize in physics (Chu, 1998; Cohen-Tannoudji, 1998; Phillips,
1998), and the pathway to optical tweezers. Single-beam trapping of microparticles
using the gradient force was finally achieved (Ashkin et al., 1986), and optical
tweezers had arrived on the scene.
The combined ability to noninvasively trap and manipulate microscopic parti-
cles and to measure forces on the order of piconewtons rapidly led to the use of
optical tweezers in a variety of biological and other fields (Block, 1992; Grier,
1997; Wright et al., 1990). The new possibility of measuring forces in microscopic
biological systems was a major advance, and allowed access to a wide variety of
measurements such as, for example, measurements of the dynamics of single DNA
molecules (Quake et al., 1997). Kuo (2001) reviews the use of optical tweezers for
measuring biological forces. Lang and Block (2003) list over a hundred papers
reporting force measurements on biological molecules, cells, or colloidal particles.
Meanwhile, parallel developments continued, for example, the demonstration
of rotational manipulation in optical tweezers by Sato et al. (1991) using the
rotation of a nonaxisymmetric trapping beam. Continuous spinning of particles
was achieved using Laguerre–Gauss mode beams which carry orbital angular
momentum about the beam axis to transfer momentum to absorbing particles
(He et al., 1995a,b). Further work led to both the controlled two-dimensional
trapping of strongly absorbing particles (Friese et al., 1998b) and the alignment
and spinning of birefringent particles without absorption (Friese et al., 1998a).
The second of these has already seen practical application (Bishop et al., 2004;
Knöner et al., 2005).
Further outgrowth of such work on rotational manipulation included the
rotation of cells and organelles (Bayoudh et al., 2003; Mohanty et al., 2004),
and progress toward optically driven micromachines (Nieminen et al., 2006).
Other interesting developments include sophisticated holographic optical twee-
zers (Dufresne et al., 2001; Grier and Roichman, 2006; Leach et al., 2004) and
optical tweezers operated remotely over the internet (Botvinick and Berns, 2005).
210 Timo A. Nieminen et al.

A. Further Reading
Ashkin, the pioneer of optical trapping and optical tweezers, has described
the development of optical tweezers in a number of papers (Ashkin, 1997, 2000).
More developments have been reviewed by Grier (2003), Molloy et al. (2003), and
Nieminen et al. (2006).
The Resource Letter by Lang and Block (2003) is an excellent starting point for
further investigation of existing work in optical tweezers.

III. Basic Principles

The optical forces that are responsible for trapping in optical tweezers result from
the transfer of momentum from the trapping beam to the particle. Recalling that
force is the time rate of change of momentum, optical forces must be accompanied
by a change in the momentum flux of the beam. Optical forces are commonly
explained in terms of the momenta of incident and reflected or refracted rays. It is
also possible to consider the angular momentum flux of the beam as a whole in
order to understand and explain optical torques.
Consider the momentum flux of a ray of light: p ¼ nP/c in the direction of the
ray, where p is the magnitude of the momentum flux, and P is the power, or energy
flux; n is the refractive index of the medium the ray is moving in, and c is the speed
of light in free space. The ratio of momentum to power for a ray is the same as for
a section of a plane wave; if we consider a plane wave to be made up of parallel
rays, we see that this must be the case. A focused beam, on the other hand, can be
considered as a bundle of converging rays.
The force exerted on a particle in optical tweezers can be understood in terms of
changes in the momentum flux of the trapping beam. For axial forces, the important
principle is that the more convergent or divergent the beam is, the lower the
momentum flux. Changes in the convergence or divergence of the beam change
the axial momentum of the beam, with a resultant axial reaction force on the
trapped particle. For radial forces, it is the direction of the beam as a whole that
needs to be considered. If the beam is deflected in one direction, it gains transverse
momentum in that direction, and the particle experiences a reaction force in the
other direction.
We can think of the trapped particles as a weak positive lens. The action of such
a lens on a beam is shown in Fig. 1. If the lens is at the focus of the beam, the rays
pass through the center of the lens and are undeviated—the optical force is zero.
If the lens is before the focus, it increases the convergence of the beam, and
therefore decreases the momentum flux. The lens gains the momentum the beam
loses, and there is a force in the direction of propagation. If the lens is after the
focus, it decreases the divergence of the beam, and hence increases the momentum
flux, resulting in a restoring force toward the focus. If the lens is displaced
sideways, the beam is deflected toward the centerline of the lens, gaining lateral
momentum. The lateral reaction force on the lens acts toward the beam axis.
6. Physics of Optical Tweezers 211

Fig. 1 The force exerted on a particle in optical tweezers can be understood in terms of changes in the
momentum flux of the trapping beam. For axial forces, the important principle is that the more
convergent or divergent the beam is, the lower the axial momentum flux. For radial forces, the direction
of the beam is the key principle. The trapped particle can be thought of as a weak positive (converging)
lens. If the lens is at the focus, the rays pass through the center of the lens and are undeviated—the
optical force is zero. If the lens is before the focus, it increases the convergence of the beam, and
therefore decreases the momentum flux. The lens gains the momentum the beam loses, and there is a
force in the direction of propagation. If the lens is after the focus, it decreases the divergence of the
beam, and hence increases the momentum flux, resulting in a restoring force toward the focus. If the lens
is displaced sideways, the beam is deflected toward the centerline of the lens, gaining lateral momentum.
The lateral reaction force on the lens acts toward the beam axis.

In addition to these forces, some rays will also be reflected, producing a force in
the direction of propagation which will displace the particle past the focus.
This simple lens model is ideal for providing a qualitative picture, but does not
provide a quantitative result for the force. First, the trapped particle interacts with
the beam in a more complex manner than a simple lens. Second, most particles
trapped in optics tweezers are too small for geometric optics to be a reliable approxi-
mation—the usual rule of thumb is that the smallest dimensions of the particle should
be at least 10 times the wavelength of the light, preferably 20l or more. Third,
trapping occurs in the vicinity of the focus of the beam, where geometric optics
fails. Despite this, many authors (Ashkin, 1992) have used the geometric optics
approximation to calculate optical forces. Wright et al. (1993, 1994) showed that
ray optics calculations are useful for large spheres.
An alternative approximate method is to consider particles that are small
compared to the wavelength—the Rayleigh limit. Despite most particles trapped
in optical tweezers being outside the size range where this approximation is valid,
this is still a useful exercise since it provides quantitative formulas which are useful
for small particles, and perhaps more importantly, have a simple physical inter-
pretation that illuminates the forces that act on larger particles. One of the eVects
of an electromagnetic field in matter to is to induce a dielectric polarization,
determined by the field and the permittivity of the medium. For optical problems,
the refractive index is determined by the permittivity [c/n ¼ (em)1/2 or c/n ¼ (em0)1/2 for
nonmagnetic materials], so the polarization is a function of the optical properties
of the medium. For the case of a sphere in a uniform field—a classic canonical
problem encountered in electromagnetics courses—a simple analytical result exists:
a dipole moment of
212 Timo A. Nieminen et al.


m2
1
d¼ 4pn2med e0 r3 E ð1Þ
m2 þ 2

is induced in the sphere by an electric field E, where nmed is the refractive index of
the surrounding medium, m is the relative refractive index of the particle, with
m ¼ npart/nmed, and we use d for the dipole moment instead of the more usual p to
avoid confusion with the symbols we use for momentum and power. From the
energy of dipole in an applied field, we can show that the force acting on the sphere
is (Harada and Asakura, 1996)

2 
m
1
Fgrad ¼ pn2med e0 r3 rjEj2
m2 þ 2

 ð2Þ
2pnmed r3 m2
1
¼ rI
c m2 þ 2

Since this is proportional to the gradient of the irradiance I, it is called the

gradient force. Note that it is also proportional to the volume of the sphere (i.e.,
proportional to r3). If a trapped particle is much smaller than the wavelength of
the trapping beam, it will be in an approximately uniform field, and Eq. (2) is a
useful approximation for the force.
If the field were static, Eq. (2) would give the total force. However, the field is time-
varying, with the direction of the induced dipole moment changing twice every
optical period. This time-varying dipole moment is equivalent to a current, and
since a magnetic field—in this case, the magnetic field of the trapping beam—exerts
a force on a current, one expects a force to act on the sphere. However, for a lossless
(i.e., nonabsorbing) material, one might reasonably expect the dipole moment to be
exactly in phase with the incident electric field, and hence the current to be a quarter
wave out of phase with the incident wave. Inconveniently for this simple picture, this
would result in a time-averaged force of zero. However, since we know that light
reflected from the particle will result in a force, this cannot be correct. In circuit theory
terms, a current quarter wave out of phase is a pure reactance, with no energy losses
involved. An oscillating dipole moment, on the other hand, is the classic ideal short
dipole antenna, which radiates energy. Since energy is removed from the incident
beam and reradiated, momentum must also be removed from the incident beam. For
a Rayleigh scatterer, the reradiation is symmetric and has a total momentum flux of
zero. Hence the momentum transfer to the particle is equal to the momentum
associated with the incident energy lost, and the force has a magnitude of (Harada
and Asakura, 1996)


8pnmed k 4 r6 m2
1
Fscat ¼ I ð3Þ
c m2 þ 2
6. Physics of Optical Tweezers 213

where k is the wavenumber of the trapping beam. This force is usually called the
scattering force. The scattering force can also be included in the expression
for the gradient force by a correction term to the polarizability, which is the ratio
of the dipole moment to the applied field such that d ¼ aE, gives an eVective
polarizability (Draine, 1988),
a0
aeff ¼ ð4Þ
1
ð2ik3 a0 Þ=3
in terms of the original uncorrected polarizability a0, making the polarizability of the
sphere complex. This has a similar eVect to the sphere possessing a complex refractive
index (i.e., the sphere being absorbing), in that it results in a force in the direction of
the incident energy flux. There is one important diVerence, however, between a
reradiating sphere and an absorbing sphere: the reradiating sphere experiences a
scattering force proportional to the square of the volume (i.e., proportional to r6),
while the absorption force on a small absorbing sphere is proportional to the volume
(so proportional to r3). If a sphere is small enough, the scattering force will always be
smaller than the gradient force, allowing, for example, highly reflective metallic
spheres to be trapped. Absorption forces, on the other hand, scale in the same way
as the gradient force, so small sizes are insuYcient to allow the trapping of absorbing
particles—the only mechanism by which suYciently absorbing particles can be
trapped in single-beam traps is to two-dimensionally trap them against a surface by
using this absorption radiation pressure force (Rubinsztein-Dunlop et al., 1998).
While the scaling of gradient forces and scattering forces with volume suggests
that, given sufficiently small sizes, any nonabsorbing particles can be three-
dimensionally trapped, this is not the case in practice. The trapped particle under-
goes random motion due to thermal fluctuations—Brownian motion. The thermal
kinetic energy of a particle in the trap is kBT, where kB is Boltzmann’s constant and
T is the absolute temperature. Noting that a conservative force is the gradient of a
scalar function and that the gradient force is proportional to the irradiance—a
scalar function—we see that the gradient force is conservative, and is the gradient of
the trapping potential


2pnmed r3 m2
1
U¼ I ð5Þ
c m2 þ 2
If this trapping potential exceeds the thermal kinetic energy significantly, the
particle is unlikely to escape due to Brownian motion. We can also compare
the trapping force against the average magnitude of the force due to Brownian
motion, which is 12pr
kBT where
is the viscosity of the fluid. Viscous drag in the
medium will assist the trap in preventing escape; consideration of this is essentially
equivalent to comparing the trapping potential with the thermal kinetic energy. In
principle, the Brownian motion force, which is independent of the trapping power,
can be overcome by the use of suYcient power, but the power required may well
destroy the particle (or even one’s optical system) in practice. The ultimate limit in the
214 Timo A. Nieminen et al.

trapping of small particles, atom trapping, is achieved by the trapping of atoms in

vacuum, or of a suYciently cold group of atoms—the Brownian motion forces
decrease with decreasing temperature.
Both the geometric optics and Rayleigh approximation provide easy to under-
stand qualitative pictures of trapping, and useful quantitative results in their respec-
tive size regimes. However, many particles in optical trapping lie between these two
regimes, and require the use of exact electromagnetic theory to calculate forces and
torques. While the mathematical formulation of the problem presents a somewhat
daunting barrier to entry, the practical implementation, especially for the case of
spherical particles, is quite feasible, and the size range for which such methods are
practical overlaps the size ranges for which the geometric optics and Rayleigh
approximations are valid. We discuss such methods later in this chapter. For now,
we mention that exact electromagnetic methods can yield agreement with experi-
ment to better than within 1% (Knöner et al., 2006). Overall, we recommend that
some caution be used when applying results obtained using approximate methods,
especially when a high level of accuracy is required.
We now proceed to present some results of applying exact electromagnetic theory
to optical trapping. Since the primary component of an optical tweezers trap is a
tightly focused laser beam, we first consider such beams in the context of exact
electromagnetic theory. We show the equivalents of some of the above considera-
tions using the exact theory. We next give some results for a representative of
trapping of a sphere, and examine the limits of trappability—what types of particles
of what sizes can be trapped in a typical optical tweezers apparatus.
Superficially, a tightly focused laser beam appears very much like a paraxial
Gaussian beam with a small beam waist radius. However, if we attempt to more
tightly focus the beam, the beam waist reaches a limiting size, while the theoretical
paraxial Gaussian waist radius can be made as small as possible. This has been
illustrated by Nieminen et al. (2003b). Furthermore, for a linearly polarized beam,
the rotational symmetry of the intensity is broken as the waist radius approaches
the limit. Components of the electric field in the direction of the beam propaga-
tion—which are assumed to be zero in the paraxial limit—approach one-third the
magnitude of the transverse components. A typical cross section of the focal plane
intensity of a tightly focused Gaussian beam is shown in Fig. 2; we also show
the significant axial component of the electric field. For a plane polarized
beam, the axial electric field in each of the two regions is p out of phase relative to
the other. Interestingly, for a circularly polarized beam, the axial component forms
an optical vortex, carrying orbital angular momentum (Nieminen et al., 2004a).
A snapshot view of the instantaneous electric field, as opposed to the electric
field amplitude, is shown in Fig. 3. The relationship between the focal volume and
the wavelength of the beam is clearly visible.
The eVect of a particle on the beam is shown in Fig. 4. The exact electromag-
netic calculations reproduce the same qualitative eVects discussed earlier, where
changes in the convergence or divergence and direction of the trapping beam result
in restoring forces acting on the particle. The force is shown in Fig. 5, as a function of
position. Note that the radial force, as a function of radial transverse displacement,
6. Physics of Optical Tweezers 215

0.4

0.3

|E |2 (a.u.)
0.2

0.1

0
1
1
0
0
x (l)
−1 −1 y (l)

0.4

0.3
|Ez|2 (a.u.)

0.2

0.1

0
1
1
0
0
x (l)
y (l)
−1 −1

Fig. 2 Irradiance distribution in the focal plane of a tightly focused Gaussian beam is shown in the left
hand plot. This beam is polarized in the x-direction, with a small but noticeable elongation of the focal
spot in this direction. This results from an axial component of the electric field caused by the focusing of
the beam. This component of the electric field parallel to the beam axis is shown in the right hand plot.
Unsurprisingly, trap stiVnesses vary between the directions parallel to and perpendicular to the plane of
polarization. While there are also small components of the electric field perpendicular to the plane
of polarization, these can be safely ignored.

is symmetric, while the axial force is not symmetric as a function of axial displace-
ment, due to the scattering force always acting in the direction of propagation of the
beam. This results in the equilibrium position of any spherical particle in the trap
being with its center beyond the focus of the beam. In most trapping situations, the
216 Timo A. Nieminen et al.

z (l)
0

−1

−2

−3
−2 −1 0 1 2
x (l)

Fig. 3 A side view of the trapping beam: the magnitude of the instantaneous electric field is shown by
the contours in the figure. The size of the focal volume compared to the wavelength of the light is clearly
evident.

weight of the trapped particle is largely compensated by buoyancy forces, and can
be neglected.
For the situation depicted in Fig. 5, the axial force is negative for a range of
distances beyond the focus. This is the force responsible for axial trapping.
An axial equilibrium position only exists if this force curve crosses the Fz ¼ 0 axis,
with a stable equilibrium in the position indicated in Fig. 5. Furthermore, the
maximum force in this reverse direction is much smaller than either the maximum
force in the forward direction or the maximum radial force. This is the usual case
for optical tweezers. Therefore, calculation of the axial force as a function of axial
displacement allows us to determine whether trapping occurs, and how strongly
the particle is trapped. For a given optical tweezers setup, it is possible to calculate
the trappability of and maximum reverse axial force acting on particles as a
function of particle refractive index and size. In Fig. 6, we show a contour map
of the maximum reverse axial restoring force as a function of size and refractive
index. Where contours are absent, there is no reverse restoring force—the axial
force is always in the direction of propagation, and the particle cannot be trapped.
For higher refractive indices, reflection of the trapping beam becomes more impor-
tant as the axial scattering force begins to push the particle out of the trap. The
reflectivity of a thin layer varies periodically with thickness depending on whether
6. Physics of Optical Tweezers 217

−2
0.12
0.14
−2
−1
0.1 0.12
−1
0 0.1
Position Z [l]

0.08

Position Z [l]
0
0.08
1 0.06
1 0.06
2 0.04
0.04
2
3 0.02
0.02
3
−2 −1 0 1 2 Intensity −2 −1 0 1 2 Intensity
Position X [l] Position X [l]
× 10−3 × 10−3
8 8
With particle With particle
7 Without particle 7 Without particle

6
6
Field intensity

Field intensity

5
5
4
4
3
3
2
2 1

1 0
−3 −2 −1 0 1 2 3 −2.0 −1.5 −1.0 −0.5 0 0.5 1.0 1.5 2.0
Position x (l) Position x (l)

Fig. 4 Change in convergence and direction of the trapping beam due to a particle in the trap
displaced away from the equilibrium position. On the left, the eVect of a particle displaced along the
axis of a downward-propagating beam below the equilibrium position. The contours in the upper plot
show the intensity, while the lower graph shows the fields 2 mm below the focus. It can be seen that the
particle acts to reduce the divergence of the beam, with an increase in the momentum flux of the beam,
resulting in an upward restoring force. On the right, the eVect of a particle displaced to the side of the
equilibrium position is shown. Note that the beam is deflected to the right, with a resulting reaction force
on the particle acting to the left.

light reflected from the front and back interfaces interferes constructively or de-
structively, and we can see that microspheres show similar behavior. This results in
the variation in trappability with size seen in the upper region in the figure. While
high-index particles such as diamond (indicated by the dashed line in the figure) are
218 Timo A. Nieminen et al.

0.14

0.12

0.10

0.08

Force in z-direction
Particle in front Particle behind
0.06 of beam waist beam waist

0.04

0.02
Equilibrium
0

−0.02

−0.04

−0.06
−1.5 −1.0 −0.5 0 0.5 1.0 1.5 2.0 2.5
z-Position of particle (l)

0.20

0.15

0.10
Force in x-direction

0.05

−0.05

−0.10

−0.15

−0.20
−1.5 −1.0 −0.5 0 0.5 1.0 1.5
x-Position of particle (l)

Fig. 5 Axial and radial forces. Note that the axial force is asymmetric. This is due to the contribution
of the scattering force.

often considered untrappable, weak trapping should be possible for specific sizes.
At the far left, we enter the Rayleigh regime, where the gradient force is propor-
tional to r3, while the scattering force is proportional to r6. As a result, the trapping
forces become small, but even high-index particles can be trapped as long as
Brownian forces can be overcome.
6. Physics of Optical Tweezers 219

2.6

2.4

Particle refractive index

2.2

2.0

1.8

1.6

1.4

0.5 1.0 1.5 2.0 2.5 3.0

Particle diameter (µm)

Fig. 6 Map of maximum upward axial restoring force as a function of size and refractive index. These
results are for trapping in water at a free-space wavelength of 1064 nm focused by an NA ¼ 1:3
microscope objective. Polystyrene microspheres (n ¼ 1:55, shown by the dotted line) can be readily
trapped. From this map, 1-mm polystyrene spheres are more weakly trapped than 2-mm spheres, as
observed experimentally. For higher refractive indices, reflections become more important as the axial
scattering force begins to push the particle out of the trap. The reflectivity of a thin layer varies with
thickness depending on whether light reflected from the front and back interfaces interfere constructively
or destructively. Similar behavior is seen for microspheres, resulting in the variation in trappability with
size in the upper region in the figure. While high-index particles such as diamond (dashed line) are often
considered untrappable, weak trapping should be possible for specific sizes. At the far left, we enter the
Rayleigh regime, where the gradient force is proportional to r3, while the scattering force is proportional
to r6. As a result, the trapping forces become small, but even high-index particles can be trapped.

A. Further Reading
The picture of a trapped particle as a lens changing the convergence or divergence
and direction of the trapping beam was suggested by Ashkin et al. (1986). Ashkin
(1992) also provided the classic quantitative treatment of optical force in terms of
rays. Geometric optics has also been used for the calculation of forces and torques
on nonspherical particles (Gauthier, 1997; Ukita and Nagatomi, 1997).
The use of the Rayleigh approximation to provide simple expressions for the
gradient force and the scattering force appears in much of the literature (Ashkin
et al., 1986). The optical force acting on Rayleigh particles is well treated by Harada
and Asakura (1996) and Chaumet and Nieto-Vesperinas (2000), the former using SI
units (i.e., MKS), the latter Gaussian units (i.e., cgs). Harada and Asakura (1996),
in particular, investigate the bounds of validity of the Rayleigh approximation for
determining optical forces and show that it is accurate for particles of radius <0.1l.
An earlier study of the trapping of Rayleigh particles was undertaken by Visscher
and BrakenhoV (1992).
220 Timo A. Nieminen et al.

IV. Fundamental Theory

Here we review some of the fundamental theory underlying optical tweezers.

This includes the transport of energy, momentum, and angular momentum by
monochromatic electromagnetic waves, and the transfer of these quantities to a
particle. We do not attempt to give thorough and detailed derivations—these are
readily available in the literature. Instead, we give the most important results, and
explain the underlying physics. We also highlight some points that are often neglected
in the literature or are points of contention.
The theory behind optical tweezers is essentially classical electromagnetic theo-
ry, so we present a brief review of the relevant basic elements of electromagnetic
theory, largely as a reminder about basic concepts and terminology. After a general
introduction, we will restrict ourselves to monochromatic electromagnetic waves in
source-free lossless (i.e., nonabsorbing) homogeneous linear media, and we will
neglect dispersion. The results we present should not uncritically be taken as being
general outside these conditions. We hope that the physical explanations we oVer
will give an indication that a result might be more general.
The interaction between electric and magnetic fields (the electric intensity E,
electric displacement D, magnetic intensity H, and magnetic induction B) and their
sources—the charge density r and the current density J—is described by the
Maxwell equations:
@B
rEþ ¼0 ð6Þ
@t

@D
rH
¼J ð7Þ
@t

rD ¼ r ð8Þ

rB ¼ 0 ð9Þ
The Lorentz force law,
f ¼ rE þ J  B ð10Þ
giving the force per unit volume, can be taken to be an operational definition of
E and B.
The properties of a material medium are contained in the constitutive relations:
D ¼ eE ð11Þ

B ¼ mH ð12Þ

J ¼ sE ð13Þ
6. Physics of Optical Tweezers 221

The last of these is Ohm’s law. In free space (i.e., vacuum), e and m become
e0 and m0. Since D and H are defined in terms of the sources of the fields (the
charge density r and the current density J ), e0 and m0 can be thought of as unit
conversion factors. A system of units can be chosen so that they are both
dimensionless and numerically equal to one; but we will use SI units throughout,
where this is not the case.
The real (i.e., not complex) fields in the Maxwell equations are functions of both
time and position. If the time-variation is sinusoidal, we can eliminate the time
derivatives by adopting the use of complex field amplitudes, reducing the system
of equations to containing functions of position only. We replace the field vari-
ables in the equations above by complex equivalents, using relations of the form
 
Eactual ¼ Re Ecomplex expð
iotÞ ð14Þ
Since the Maxwell equations must hold for all times, they must also hold for the
imaginary part of the complex field, so there is no need to explicitly take the real
part in the equations. The complex amplitudes Ecomplex, and so on, are indepen-
dent of time, so the time derivatives can be written analytically. The exponential
variation with time is the same for every single term in the equation, and can
therefore be eliminated from the equations. On occasion, it is necessary to recall
that the complex amplitudes are not the instantaneous fields, and to use Eq. (14) to
obtain the real instantaneous fields.
Dropping the complex subscript for brevity, the time-harmonic Maxwell equa-
tions in source-free media become:
r  E
ioB ¼ 0 ð15Þ

r  H þ ioD ¼ 0 ð16Þ

rD ¼ 0 ð17Þ

rB ¼ 0 ð18Þ
Combining the first two of these equations, and using the constitutive relations,
we obtain the vector Helmholtz equation for both the electric and magnetic fields:

r2 E þ k2 E ¼ 0 ð19Þ

r2 H þ k2 H ¼ 0 ð20Þ
while obtaining the further restrictions on the field that r E ¼ 0 and r  H ¼ 0
from the last two equations. These conditions, that the fields are divergence-free,
are sometimes specified by stating that the fields are transverse. The wavenumber k
is equal to (em)1/2o so we can identify v ¼ (em)
1/2 ¼ c/n as the phase speed of the
wave.
222 Timo A. Nieminen et al.

We will be able to restrict ourselves to considering plane waves. For a plane

electromagnetic wave, E and H are perpendicular to each other, and also to
the direction of propagation.pEffiffi and H are also in phase with each other, and
pffiffiffi
have magnitudes related to E e ¼ H m.
The Maxwell equations as presented above are diVerential equations, and
cannot be applied where the fields or complex amplitudes of the fields are discon-
tinuous. This occurs at interfaces between media with diVerent electromagnetic
or optical properties. This requires us to use boundary conditions at the interface
to relate solutions to the Maxwell or Helmholtz equations in the two media
separated by the interface. The boundary conditions which we will need to
consider are simply that the vector components of E and H parallel to the interface
are the same on both sides of the interface.

V. Transport of Energy, Momentum, and Angular

Momentum by Light

It is of course evident to any who use lasers, have stood in sunlight, or otherwise
experienced related eVects, that electromagnetic waves carry energy. Those
who use optical tweezers are also familiar with the transport of momentum by
electromagnetic waves.
If we consider the work done on the currents within a volume by the electric
field E (if charges do not move, no work is done on them, and the magnetic field
does no work, so this is suYcient), we can show that the instantaneous energy flux
density (i.e., the power per unit area) of an electromagnetic wave is equal to the
Poynting vector

S ¼ Eactual  Hactual ð21Þ

For a monochromatic wave, especially at optical frequencies, we are usually

interested in the time-averaged Poynting vector in terms of the complex amplitudes
rather than the real fields, which is equal to

1
S ¼ E  H ð22Þ
2
where * denotes the complex conjugate. For a plane electromagnetic wave, the
Poynting vector is always in the direction of propagation of the wave, and has a
magnitude of

ce
S¼ jE j2 ð23Þ
2n
There is a close relationship between the energy flux density (i.e., the Poynting
vector), the energy density u of the wave, and the velocity v of energy transport:
6. Physics of Optical Tweezers 223

S ¼ uv ð24Þ
This is a general relationship which holds for all waves. Since the time-averaged
energy density is
1
u ¼ ðE  D þ B  H Þ ð25Þ
4
which simplifies to
e
u ¼ jE j2 ð26Þ
2
for a plane wave, we can verify that the speed of energy transport in the simple
media we are considering is equal to the phase speed of the wave, c/n. (For more
general media, for which the permittivity can be complex, this is not necessarily
the case.)
The transport of energy by a wave necessitates the transport of momentum, and
the momentum flux density p is determined by the energy flux density and the speed
of transport of energy. The momentum transported per unit area in the direction of
propagation of the wave is
p ¼ nS=c ð27Þ
The magnitude of the momentum flux is numerically equal to the energy
density, since both are related to the energy flux by the same speed. At this
point, it is necessary to mention that there is a long-standing controversy over
the momentum of an electromagnetic wave in matter—whether the momentum
flux density is nS/c, S/(nc), or some other value—often called the Abraham–
Minkowski controversy. The relationships between energy density, energy flux,
and the speed at which energy is transported are general relationships valid for
all waves, and therefore the momentum flux in Eq. (27) must be the total momen-
tum flux associated with the electromagnetic wave, and is independent of the
Abraham–Minkowski controversy. A review of the controversy (Pfeifer et al.,
2006) shows that the controversy is essentially a matter of semantics and not of
physics—there is no disagreement about the total momentum associated with an
electromagnetic wave in matter, only about the division of this total momentum
into components called electromagnetic momentum and material momentum.
Since it is the total momentum that produces physical eVects, there is no way to
physically distinguish between the diVerent expressions for the electromagnetic
momentum.
Finally, any transport of momentum implies a transport of angular momentum
as well. In the case of electromagnetic fields, this presents some diYculties. While it
is tempting to simply say that the angular momentum flux is the moment of the
linear momentum flux, that is,
S
j ¼ r  p ¼ nr  ð28Þ
c
224 Timo A. Nieminen et al.

this is not correct, as it neglects the fact that electromagnetic waves can carry
spin angular momentum. In general, the angular momentum flux can be divided
into an orbital angular momentum flux l and spin s—named in analogy with
classical mechanics where a body can both undergo orbital motion and spin about
its own axis—such that
j¼lþs ð29Þ
Spin is defined as intrinsic angular momentum—angular momentum that is
independent of the origin about which we choose to take our moments. Therefore,
spin is the portion of the angular momentum which is independent of r, and can
easily be found from an expression for the total angular momentum, even if
complicated, by choosing r to be zero. The remainder of the angular momentum
is the orbital angular momentum. The use of Eq. (28) is equivalent to stating that
electromagnetic waves cannot carry spin.
The spin flux of a plane electromagnetic wave depends on its degree of circular
polarization s and is equal to
sS
s¼ ð30Þ
o
where o is the optical angular frequency, and the degree of circular polarization is
defined as
s ¼ ðIL
IR Þ ¼ I ð31Þ
where IL and IR are the irradiances of the left- and right-circularly polarized
components (Nieminen et al., 2001a). This spin flux is equivalent to the quantum
mechanical result that photons carry Ž spin.
However, it has been known for a long time that for fields finite in extent, the total
angular momentum is given by the integral over the entire field of the moment of the
linear momentum density (Humblet, 1943), and a similar result also applies to
the angular momentum flux (Crichton and Marston, 2000), so that
ð ð
S
J ¼ r  pdA ¼ nr  dA ð32Þ
c
It is natural (although incorrect in this case) to identify the integrand as the flux
density of the total flux, and this has given rise to a controversy about the angular
momentum content of a circularly polarized plane wave. However, since we will be
dealing with physically realizable waves in practice, rather than mathematically
convenient but unphysical entities such as infinite plane waves, and there is no
doubt about the total angular momentum flux of such waves, we can safely ignore
the controversy.
Now that we have identified the energy and momentum flux densities, and a
method of obtaining the total angular momentum flux, we can consider the applica-
tion of these to the question of optical trapping. To do so we need to consider non-
plane waves—the incident beam essentially consists of spherical waves converging to
6. Physics of Optical Tweezers 225

the focus and then diverging away as spherical waves, whether refracted or reflected
or not by a particle in the trap. Our method will be to consider the total energy,
momentum, and angular momentum fluxes through a spherical surface centered on
the focus. These directly give the power absorbed by the particle, and the optical force
and torque—if, for example, the total momentum flux into the volume enclosed by
the spherical surface is nonzero, this means that the total momentum inside the
spherical surface is changing. Since force is the rate of change of momentum, this is
the optical force acting on the particle in the trap. Similarly, the total energy and
angular momentum fluxes into the enclosed sphere give the absorbed power and the
optical torque. As long as the radius of the spherical surface is much greater than the
wavelength, on any small portion of the spherical surface the waves will very closely
approximate plane waves, and we can use the plane wave formulas for the energy,
momentum, and angular momentum fluxes. We can also center the spherical surface
on a particle within the trap and obtain the same results for power, force, and torque.
Since we are using a large (compared to the wavelength) radius for the spherical
surface, the Poynting vector will be very nearly purely radial at the surface.
To calculate the absorbed power, it is simplest to proceed in spherical coordi-
nates, and we can simply write the total energy flux into the sphere in terms of the
radial component of the Poynting vector as
ð
Pabs ¼
r2 Sr dO ð33Þ

where the integration is taken over all angles. The negative sign is a result of the
desired quantity depending on the total flux into the sphere, while Sr is outward
when positive. The radius r is the radius of the spherical surface, and appears as
the factor of r2 because the surface area of the spherical surface is proportional to
r2. Note that since the irradiance of spherical waves falls oV according to an
inverse square law, the Poynting vector is proportional to 1/r2, so the r2 term
serves to keep the total flux independent of the choice of r. If we have an analytical
expression for the Poynting vector, or at least its radial component, it is usually
convenient to take the limit of r2Sr as r ! 1. If the Poynting vector is calculated
numerically, then the integration over the surface can be performed numerically.
For calculation of the optical force, it is most convenient to proceed with the
Poynting vector expressed in Cartesian coordinates, since we usually wish to know
the Cartesian components of the force. The force is given by
ð
F ¼
r2 S dO ð34Þ

Again, it can be useful to take large r limits before integrating.

The calculation of optical torque is similar in many respects, but with one
crucial diVerence. We can write the torque as
226 Timo A. Nieminen et al.

ð
t ¼
r2 r  S dO ð35Þ

but we must make sure that, if taking large r limits, that the limit of r2r  S is
taken, rather than taking the moment of the limit of r2S, which would always yield
a torque of zero.
While these formulas can be used to obtain the quantities of most interest in
optical trapping, they are of little use unless we know the fields E and H which we
require to find the Poynting vector on the spherical surface. Determining these
fields is the subject of the next section. Before passing on to this, however, we will
consider some further details of the interaction between the trapping beam and a
particle within the trap.
First, if the particle in the trap is nonabsorbing, the absorbed power will be
zero, and there will be no need to calculate the energy flux, except possibly to
check the accuracy of numerical calculation of the fields. On the other hand, if the
particle is absorbing, the absorbed power is of great interest, as, for example, it
allows us to estimate the likelihood of opticution of live specimens.
Second, if the particle is completely spherically symmetric, which requires the
particle to be spherical, homogeneous, and optically isotropic, the only possible
mechanism allowing an optical torque to be exerted on the particle is absorption.
Therefore, if one is trapping typical nonabsorbing microspheres, there will be no
need to calculate torques.
Finally, if the trapped particle is stationary, then, despite any optical force
exerted on it, the electromagnetic field does no work on the particle. On the
other hand, if there is an optical force, and the particle is moving, or if there is
an optical torque, and the particle is rotating, the trapping beam will do work on
the particle. We might then ask from where, if there is no absorption, does this
energy come from? The answer is that in these cases, there must be a Doppler shift
in the frequency. In fact, the Doppler shift due to the reflection of a plane wave
from a moving mirror can be used to obtain the energy–momentum relationship
we made use of earlier in this section. In the case of rotational motion, there is a
rotational frequency shift.

A. Further Reading
We have considerably simplified matters here by dealing with plane waves. In this
case, the direction of transport of momentum and the direction of the transported
momentum coincide. The same is true for the spin flux and the spin. This allows us
to write the momentum and angular momentum fluxes simply as vectors. If these
directions did not coincide, we would need to write these fluxes as 3  3 tensors—the
Maxwell stress tensor and the moment of the Maxwell stress tensor.
General treatments of electromagnetic theory are many, but sources specifically
dealing with the transport of energy, momentum, and angular momentum in general
are relatively few. Nonetheless, the topic of transport of energy and momentum is
covered in many textbooks (GriYths, 1999; Jackson, 1999; Stratton, 1941).
6. Physics of Optical Tweezers 227

The general principles of transport of energy in media by waves were first for-
mulated by N. A. Umov in 1874; the Poynting vector is usually called the Umov
vector or the Umov–Poynting vector in the Russian literature. Unfortunately,
Umov’s papers do not seem to be readily available in English. The transport of
energy specifically by electromagnetic fields was treated 10 years later by Poynting
(1884) and Heaviside. Poynting (1909) appears to have been the first to explicitly
consider the transport of spin angular momentum by electromagnetic waves.
Humblet (1943) is generally credited with being the first to resolve the ‘‘angular
momentum paradox,’’ but this failed to end the controversy. Debate over this issue
continues, with the fires kept well stoked by the fuel of optical rotation in laser
trapping. Recent works such as Benford and Konz (2004), Nieminen (2004),
Stewart (2005), and Zambrini and Barnett (2005) convey the nature of the modern
debate.
Further mention should be made of orbital angular momentum. Optical orbital
angular momentum is often described as being somehow special. However, any
laser beam or other electromagnetic beam possesses orbital angular momentum if a
suitable origin about which to take moments is chosen. For example, most macro-
scopic measurements of radiation pressure, such as those of Lebedew and Nichols
and Hull, have actually been measurements of orbital angular momentum—such
experiments typically make use of a torsion pendulum and measure a torque, and
hence the orbital angular momentum of the beam about the axis of the pendulum.
Since the orbital angular momentum transfer in such a case is equal to two times the
product of the lever arm of the mirror and the linear momentum of the beam,
the linear momentum can be measured.
This does not alter the fact that the orbital angular momentum of certain types
of beams, those possessing optical vortices, is special—these beams carry a total
angular momentum about the beam axis that is nonzero. Equation (28) should
give the orbital angular momentum flux, and generalizing this to the beam as a
whole, we expect that if we take moments about a point through which the beam
axis passes, the orbital angular momentum would be zero.
However, for special types of beams, we find that the orbital angular momentum
density about the beam axis is nonzero, and the total orbital angular momen-
tum density about the beam axis is also nonzero. This is a truly remarkable result,
and such beams have been discussed in a number of papers (Allen et al., 1999, 2003).
One especially interesting feature of such beam is that while the orbital angular
momentum density depends on the choice of origin about which moments are taken
(by definition), the total angular momentum about the beam axis—which would
be zero for ‘‘conventional’’ beams such as a typical Gaussian beam—is independent
on this choice (O’Neil et al., 2002).
We also note that while there is a monosyllabic technical term for ‘‘intrinsic angular
momentum’’ (i.e., ‘‘spin’’), there is currently no equivalent term for ‘‘orbital ang-
ular momentum.’’ This results in the use of, for example, ‘‘OAM,’’ which can equally
mean ‘‘optical angular momentum’’ or ‘‘orbital angular momentum.’’ We believe
that the optical angular momentum is of suYcient importance in optical trapping and
228 Timo A. Nieminen et al.

other applications to merit its own specific terminology, and recommend, and look
forward to, the adoption of the term ‘‘whorl.’’
We can also note that there are two distinct pictures of the transfer of optical
energy, momentum, and angular momentum to a particle. Above, we consider
these in terms of fluxes through a surface. We could also have considered them in
terms of the Lorentz force law, finding the work done on currents by the electro-
magnetic field to find the power, and the force and torque exerted on charge,
current, and dielectric polarization densities to find the optical force and torque.
Both approaches must yield the same results. Our flux-based approach allows the
problem to be reduced to two-dimensional integrals over a surface, while a
Lorentz force-based approach requires three-dimensional volume integrals. The
theoretical equivalence of the two approaches has been shown to hold numerically
(Kemp et al., 2005) and the Lorentz force approach has also been exploited to
determine the internal stresses within a particle (Hoekstra et al., 2001).
The transfer of angular momentum to a particle through scattering is strongly
dependent on the symmetry of the particle and the electromagnetic field. We review
this topic in Nieminen et al. (2004c).
Finally, Atkinson (1935) discusses the early work on rotational frequency shifts.
Some of the modern papers include Courtial et al. (1998) and Padgett (2004).

VI. Computational Modeling of Optical Tweezers

As noted earlier, and shown in Fig. 7, optical tweezers usually fall into the gap
between the regimes of applicability of large particle approximations such as
geometric optics and small particle approximations such as Rayleigh scattering.
This size range, where dimensions of the particle are comparable to the wavelength
of the light, is often called the resonance region. As a result, computational modeling
of optical tweezers usually requires resort to direct solution of the Maxwell equations
(primarily time-domain methods) or the vector Helmholtz equation (frequency-
domain methods). As noted earlier, the vector Helmholtz equation follows from
the Maxwell equations for the case of monochromatic illumination in linear isotropic
media, with the further restriction that the solutions must be divergence-free (i.e.,
r  E ¼ 0 and r  H ¼ 0) also following from the Maxwell equations. While special
cases such as particles with nonlinear optical response, anisotropic particles, and
trapping by short-pulse lasers might require direct time- or frequency-domain solu-
tion of the Maxwell equations, solution of the Helmholtz equation is almost always
suYcient.
For modeling the trapping of spherical particles (by which we also mean that
the particles are homogeneous and optically linear and isotropic), the existence of
the analytical Lorenz–Mie solution for scattering by a sphere means that optical
forces and torques can be calculated with relatively little eVort. However, Lorenz
and Mie dealt with the case of plane wave illumination, and optical tweezers require
6. Physics of Optical Tweezers 229

Increasing refractive index

Rayleigh−Gans

Geometric optics
Rayleigh scattering
Optical tweezers

l/2 5l
Increasing size

Fig. 7 Map of regimes of validity of some computational methods for calculating the scattering of
light by particles. Note that typical particles trapped in optical tweezers fall outside the regimes of
applicability of the approximate methods, and require direct solution of the Maxwell equations or the
vector Helmholtz equation.

a tightly focused beam. The extension of the original Lorenz–Mie theory to arbi-
trary illumination is usually called generalized Lorenz–Mie theory (GLMT).
Lorenz–Mie theory is based on the representation of the incident and scattered
electromagnetic waves as sums of vector spherical wavefunctions (VSWFs). The
incident field can be written as

1 X
X n
Einc ¼ anm Mð3Þ ð3Þ
nm þ bnm Nnm ð36Þ
n¼1 m¼
n

and the scattered field as

1 X
X n
Escat ¼ pnm Mð1Þ ð1Þ
nm þ qnm Nnm ð37Þ
n¼1 m¼
n

where n is the radial mode index, m is the azimuthal mode index, anm, bnm, pnm, and
qnm are the mode amplitudes of the incident and scattered fields (often called
expansion coeYcients or multipole coeYcients), and Mð3Þ ð3Þ ð1Þ ð1Þ
nm , Nnm ; Mnm , and Nnm are
the vector spherical wavefunctions of the third and first type, respectively. While
the above expressions for the fields involve infinite sums, the sum for the scattered
field produced by a sphere of radius r can be truncated at a finite nmax approxi-
mately equal to the size parameter (kr, the product of the wavenumber and the
radius). For spheres on the order of the wavelength in size, only a modest number
of terms are required. Where the illumination is also finite in width, the sum for
the incident field also converges similarly.
230 Timo A. Nieminen et al.

The orthogonality properties of the vector spherical wavefunctions—the behav-

ior of products of VSWFs integrated over a spherical surface—lead to two impor-
tant results. First, there is no coupling between diVerent modes when the scatterer is
a homogeneous isotropic sphere. The scattered mode amplitudes can be simply
found from the incident wave mode amplitudes, with

pnm ¼ an anm ð38Þ

qnm ¼ bn bnm ð39Þ

where an and bn are the Mie scattering coeYcients. This is the key result of
Lorenz–Mie theory.
The second result following from the orthogonality properties of the VSWFs is
a simplification of the calculation of the optical force and torque. Recalling that
the absorbed power, force, and torque can be found by integration of the Poynting
vector over a spherical surface surrounding the particle [Eqs. (34) and (35)], we
might at first expect that it is necessary to perform these integrations numerically
by calculating the electric and magnetic fields over a grid of points on a sphere,
which would be a computationally intensive task. However, if one substitutes
expressions (36) and (37) for the fields in this calculation, the resulting formulas
involve integrals of products of VSWFs, the values of which are known from the
orthogonality properties. Therefore, formulas for the absorbed power, force, and
torque involving sums over the mode amplitudes and their products result, which
can be readily calculated. A thorough treatment of this is given by Farsund and
Felderhof (1996). Unfortunately, there is no universal normalization or sign and
parity conventions for the VSWFs, and care is required to check that the formulas
used for power, force, and torque calculations match the VSWFs used.
In the past, the practical application of Lorenz–Mie theory was highly labor
intensive—calculation of the Mie coeYcients requires the calculation of Ricatti–
Bessel functions or spherical Bessel and Hankel functions and their derivatives,
while calculation of the field (which requires evaluating the VSWFs) adds the
calculation of spherical harmonics and their angular derivatives. The advent of
the modern computer, and just as importantly, routines or packages that can cal-
culate the special functions required, has greatly eased what was once an onerous
computational burden.
The one remaining requirement is the calculation of the mode expansion coeY-
cients of the trapping beam. For modeling the trapping of spherical particles, this
is the most theoretically diYcult and computationally intensive component of the
overall task. While this is a far from trivial task, there are a number of ways in
which it can be done. It is possible to perform an integral transform (Maia Neto
and Nussenzweig, 2000; Mazolli et al., 2003). This is well suited to trapping by a
Gaussian beam. A very fast approximate method is the localized approximation
(Gouesbet, 1996, 1999; Gouesbet et al., 1995; Lock, 2004a,b). Our method of
choice uses an overdetermined point-matching scheme (Nieminen et al., 2003b)
6. Physics of Optical Tweezers 231

which is suitable for arbitrary beams and provides stable and robust numerical
performance and convergence.
Lorenz–Mie theory is only applicable to homogeneous isotropic spheres, and
while such spheres are often the particle of choice in optical trapping, other types
of particles are also trapped. For example, living cells often fail to accommodate
our desire for sphericity. A number of approaches are possible. Most simply, we
can assume that we can use an equivalent sphere to approximately model the
trapped object. A slightly more sophisticated approach is to model suitable objects
as layered spheres, for example a cell as a spherical nucleus surround by a
spherical layer of cytoplasm—the basic Lorenz–Mie solution is readily extensible
to concentric layered spheres. However, the rotational symmetry of a trapped
object strongly aVects the optical torque exerted on it. If we are interested in
rotational manipulation, spherical approximations completely fail us. In addition,
equivalent spheres will not give the correct optical force (Nieminen et al., 2001b)
so are not useful beyond obtaining estimates of the force. Thus, it is sometimes
necessary to resort to more general methods of calculating scattering.
In principle, any method of solution that is valid in the resonance region will
suYce. However, in practice, the requirements of modeling optical tweezers restrict
the range of methods that are feasible. In particular, one typically wants to be able
to calculate the force acting on the trapped particle at diVerent positions within the
trap, and even when simply trying to find the equilibrium position of the particle in
the trap along the beam axis, dozens of calculations are required. In calculating
three-dimensional maps of the force acting on the particle, the number of calcula-
tions can easily exceed a thousand. This requirement for eYciency is probably the
reason that so little modeling of optical tweezers has used general methods such as
the finite-diVerence time-domain method (FDTD).
Fortunately, a method is available that meets these requirements for eYciency.
Recall that the orthogonality of the VSWFs means than there is no coupling
between modes for a spherical scatterer, resulting in each scattered mode being
given directly by the incident mode and the relevant Mie coeYcient. This means
that the Mie coeYcients completely describe the scattering properties of the sphere
at that wavelength, independently of the details of the illumination. As noted
above, once the Mie coeYcients are found, they can be used for repeated calcula-
tions. For the more general case of a scatterer with linear optical properties, but
otherwise arbitrary, expressions (36) and (37) are still valid. Taking coupling
between modes into account, the equivalent of Eqs. (38) and (39) can be written
for a general scatterer:
nmax X
X n
pn 0 m 0 ¼ An0 m0 nm anm þ Bn0 m0 nm bnm ð40Þ
n¼1 m¼
n

nmax X
X n
qn 0 m 0 ¼ Cn0 m0 nm anm þ Dn0 m0 nm bnm ð41Þ
n¼1 m¼
n
232 Timo A. Nieminen et al.

where the infinite sums have been truncated at a finite nmax. If we write the incident
and scattered mode amplitudes as column vectors a and p, the above expressions
for the scattered field can be written compactly as a matrix–vector product:
p ¼ Ta ð42Þ
The matrix T is called the transition matrix, or simply the T-matrix, and just as
the Mie coeYcients are independent of the illumination, so is the T-matrix.
Therefore, if one can calculate the T-matrix for a trapped object, it can be used
repeatedly for calculating the optical force and torque for diVerent positions or
orientations within the trap.
Calculation of the T-matrix is significantly more involved than calculation of
the Mie coeYcients, but fast methods are available for homogeneous rotationally
symmetric particles, which includes a large number of shapes of interest. For
example, we have used the T-matrix method to model the optical rotation of
glass cylinders in optical tweezers (Bishop et al., 2003) and the alignment and
rotational manipulation of chloroplasts (Bayoudh et al., 2003).
While Fig. 7 suggests that, especially for biological particles which often have a
relative refractive index (m ¼ nparticle/nmedium) close to 1, methods such as anoma-
lous diVraction or other methods valid for low-contrast scatterers could be used,
these methods do not appear to have been used for modeling optical tweezers.
This may result from having to deal with the highly nonparaxial illumination
required for optical tweezers, while such methods are almost always implemented
for the calculation of scattering of plane waves. This is also, understandably, the
case for almost all computational modeling of scattering, both in the resonance
region or otherwise. While many light and electromagnetic scattering computer
codes are available, one should not expect to be able to use them for the modeling
of optical tweezers without a significant investment in modification of the original
codes to accommodate the focused beam.

A. Further Reading
The classic treatment of Lorenz–Mie theory is by van de Hulst (1957). The topic
is briefly covered in many books on electromagnetic theory, for example, by
Stratton (1941) and Jackson (1999). The subject of scattering by spheres has
long been of interest, and the early history of the field is reviewed by Logan
(1965). The original papers were by Lorenz (1890) and Mie (1908).
The application of Lorenz–Mie theory to modeling the optical trapping of
spheres in Gaussian beams has been comprehensively covered by Mazolli et al.
(2003), following an earlier, shorter, paper by Maia Neto and Nussenzweig (2000).
Earlier modeling of trapping using Mie theory includes the work of Ren
et al. (1996).
The T-matrix method was first introduced by Waterman (1965, 1971). A widely
used T-matrix code was developed by Michael Mishchenko, and is publicly avail-
able at http://www.giss.nasa.gov/crmim/. A variety of light scattering computer
6. Physics of Optical Tweezers 233

codes, and other resources for electromagnetic and light scattering, are available at
http://www.t-matrix.de/.
Our own computational methods for modeling optical tweezers are described in
Nieminen et al. (2003a,b, 2004b).

References
Allen, L., Barnett, S. M., and Padgett, M. J. (eds.) (2003). ‘‘Optical Angular Momentum.’’ IOP, Bristol.
Allen, L., Padgett, M. J., and Babiker, M. (1999). The orbital angular momentum of light. Prog. Opt.
39, 291–372.
Ashkin, A. (1970). Acceleration and trapping of particles by radiation pressure. Phys. Rev. Lett. 24,
156–159.
Ashkin, A. (1992). Forces of a single-beam gradient laser trap on a dielectric sphere in the ray optics
regime. Biophys. J. 61, 569–582.
Ashkin, A. (1997). Optical trapping and manipulation of neutral particles using lasers. Proc. Natl. Acad.
Sci. USA 94, 4853–4860.
Ashkin, A. (2000). History of optical trapping and manipulation of small-neutral particle, atoms, and
molecules. IEEE J. Select. Top. Quantum Electron. 6, 841–856.
Ashkin, A., Dziedzic, J. M., Bjorkholm, J. E., and Chu, S. (1986). Observation of a single-beam gradient
force optical trap for dielectric particles. Opt. Lett. 11, 288–290.
Atkinson, R.d’E. (1935). Energy and angular momentum in certain optical problems. Phys. Rev. 47,
623–627.
Bayoudh, S., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2003). Orientation of
biological cells using plane polarised Gaussian beam optical tweezers. J. Mod. Opt. 50, 1581–1590.
Benford, G., and Konz, C. (2004). Reply to comment on ‘‘Geometric absorption of electromagnetic
angular momentum.’’ Opt. Commun. 235, 231–232.
Bishop, A. I., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2003). Optical
application and measurement of torque on microparticles of isotropic nonabsorbing material.
Phys. Rev. A 68, 033802.
Bishop, A. I., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2004). Optical
microrheology using rotating laser-trapped particles. Phys. Rev. Lett. 92, 198104.
Block, S. M. (1992). Making light work with optical tweezers. Nature 360, 493–495.
Botvinick, E. L., and Berns, M. W. (2005). Internet-based robotic laser scissors and tweezers micro-
scopy. Microsc. Res. Tech. 68, 65–74.
Chaumet, P. C., and Nieto-Vesperinas, M. (2000). Time-averaged total force on a dipolar sphere in an
electromagnetic field. Opt. Lett. 25, 1065–1067.
Chu, S. (1998). The manipulation of neutral particles. Rev. Mod. Phys. 70, 685–706.
Cohen-Tannoudji, C. N. (1998). Manipulating atoms with photons. Rev. Mod. Phys. 70, 707–719.
Courtial, J., Robertson, D. A., Dholakia, K., Allen, L., and Padgett, M. J. (1998). Rotational frequency
shift of a light beam. Phys. Rev. Lett. 81, 4828–4830.
Crichton, J. H., and Marston, P. L. (2000). The measurable distinction between the spin and orbital
angular momenta of electromagnetic radiation. Electron. J. DiVer. Eq., Conf. 4, 37–50.
Draine, B. T. (1988). The discrete-dipole approximation and its application to interstellar graphite
grains. Astrophys. J. 333, 848–872.
Dufresne, E. R., Spalding, G. C., Dearing, M. T., Sheets, S. A., and Grier, D. G. (2001). Computer-
generated holographic optical tweezer arrays. Rev. Sci. Instrum. 72, 1810–1816.
Farsund, ., and Felderhof, B. U. (1996). Force, torque, and absorbed energy for a body of arbitrary
shape and constitution in an electromagnetic radiation field. Phys. A 227, 108–130.
Friese, M. E. J., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (1998a). Optical
alignment and spinning of laser trapped microscopic particles. Nature 394, 348–350; [Erratum in
Nature 395, 621.
234 Timo A. Nieminen et al.

Friese, M. E. J., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (1998b). Optical
torque controlled by elliptical polarization. Opt. Lett. 23, 1–3.
Gauthier, R. C. (1997). Theoretical investigation of the optical trapping force and torque on cylindrical
micro-objects. J. Opt. Soc. Am. B 14, 3323–3333.
Gouesbet, G. (1996). Exact description of arbitrary-shaped beams for use in light-scattering theories.
J. Opt. Soc. Am. A 13, 2434–2440.
Gouesbet, G. (1999). Validity of the localized approximation for arbitrary shaped beams in the
generalized Lorenz-Mie theory for spheres. J. Opt. Soc. Am. A 16, 1641–1650.
Gouesbet, G., Lock, J. A., and Gréhan, G. (1995). Partial-wave representations of laser beams for use in
light-scattering calculations. Appl. Opt. 34, 2133–2143.
Grier, D. G. (1997). Optical tweezers in colloid and interface science. Curr. Opin. Coll. Interface Sci. 2,
264–271.
Grier, D. G. (2003). A revolution in optical manipulation. Nature 424, 810–816.
Grier, D. G., and Roichman, Y. (2006). Holographic optical trapping. Appl. Opt. 45, 880–887.
GriYths, D. J. (1999). ‘‘Introduction to Electrodynamics,’’ 3rd edn. Prentice Hall, Upper Saddle
River, NJ.
Harada, Y., and Asakura, T. (1996). Radiation forces on a dielectric sphere in the Rayleigh scattering
regime. Opt. Commun. 124, 529–541.
Hardin, C. L. (1966). The scientific work of the Reverend John Michell. Ann. Sci. 22, 27–47.
He, H., Friese, M. E. J., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (1995a). Direct observation of
transfer of angular momentum to absorptive particles from a laser beam with a phase singularity.
Phys. Rev. Lett. 75, 826–829.
He, H., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (1995b). Optical particle trapping with higher-
order doughnut beams produced using high eYciency computer generated holograms. J. Mod. Opt.
42, 217–223.
Hoekstra, A. G., Frijlink, M., Waters, L. B. F. M., and Sloot, P. M. A. (2001). Radiation forces in the
discrete-dipole approximation. J. Opt. Soc. Am. A 18, 1944–1953.
Humblet, J. (1943). Sur le moment d’impulsion d’une onde électromagnétique. Physica 10, 585–603.
Jackson, J. D. (1999). ‘‘Classical Electrodynamics,’’ 3rd edn. John Wiley, New York.
Kemp, B. A., Grzegorczyk, T. M., and Kong, J. A. (2005). Ab initio study of the radiation pressure on
dielectric and magnetic media. Opt. Express 13, 9280–9291.
Keplero, J. (1619). ‘‘De cometis libelli tres,’’ Augustae Vindelicorum, Augsburg (in Latin).
Knöner, G., Parkin, S., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2005). Microrheology using
dual beam optical tweezers and ultrasensitive force measurements. Proc. SPIE 5736, 73–80.
Knöner, G., Parkin, S., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2006).
Measurement of refractive index of single microparticles. Phys. Rev. Lett. 97, 157402.
Kuo, S. C. (2001). Using optics to measure biological forces and mechanics. TraYc 2, 757–763.
Lang, M. J., and Block, S. M. (2003). Resource Letter: LBOT-1: Laser-based optical tweezers. Am.
J. Phys. 71, 201–215.
Leach, J., Sinclair, G., Jordan, P., Courtial, J., Padgett, M. J., Cooper, J., and Laczik, Z. J. (2004). 3D
manipulation of particles into crystal structures using holographic optical tweezers. Opt. Express 12,
220–226.
Ledebew, P. (1902). An experimental investigation of the pressure of light. Astrophys. J. 15, 60–62.
Lock, J. A. (2004a). Calculation of the radiation trapping force for laser tweezers by use of generalized
Lorenz–Mie theory. I. Localized model description of an on-axis tightly focused laser beam with
spherical aberration. Appl. Opt. 43, 2532–2544.
Lock, J. A. (2004b). Calculation of the radiation trapping force for laser tweezers by use of generalized
Lorenz–Mie theory. II. On-axis trapping force. Appl. Opt. 43, 2545–2554.
Logan, N. A. (1965). Survey of some early studies of the scattering of plane waves by a sphere. Proc.
IEEE 53, 773–785.
Lorenz, L. (1890). Lysbevægelsen i og uden for en af plane Lysblger belyst Kugle. Videnskabernes
Selskabs Skrifter 6, 2–62.
6. Physics of Optical Tweezers 235

Maia Neto, P. A., and Nussenzweig, H. M. (2000). Theory of optical tweezers. Europhys. Lett. 50,
702–708.
Maxwell, J. C. (1865). A dynamical theory of the electromagnetic field. Philos. Trans. R. Soc. Lond. 155,
459–512.
Maxwell, J. C. (1873). ‘‘A Treatise on Electricity and Magnetism.’’ (in two volumes) Clarendon Press,
Oxford.
Mazolli, A., Maia Neto, P. A., and Nussenzveig, H. M. (2003). Theory of trapping forces in optical
tweezers. Proc. R. Soc. Lond. A 459, 3021–3041.
Mie, G. (1908). Beiträage zur Optik trüber Medien, speziell kolloidaler Metallösungen. Annalen der
Physik 25, 377–445.
Mohanty, S. K., Uppal, A., and Gupta, P. K. (2004). Self-rotation of red blood cells in optical tweezers:
Prospects for high throughput malaria diagnosis. Biotechnol. Lett. 26, 971–974.
Molloy, J. E., Dholakia, K., and Padgett, M. J. (2003). Optical tweezers in a new light. J. Mod. Opt. 50,
1501–1507. (preface to special issue).
Nichols, E. F., and Hull, G. F. (1901). A preliminary communication on the pressure of heat and light
radiation. Phys. Rev. 13, 307–320.
Nichols, E. F., and Hull, G. F. (1902). Pressure due to light and heat radiation. Astrophys. J. 15, 62–65.
Nichols, E. F., and Hull, G. F. (1903a). The pressure due to radiation. Phys. Rev. 17, 26–50, 91–104.
Nichols, E. F., and Hull, G. F. (1903b). The pressure due to radiation. Astrophys. J. 17, 315–351.
Nieminen, T. A. (2004). Comment on ‘‘Geometric absorption of electromagnetic angular momentum,’’
C. Konz, G. Benford. Opt. Commun. 235, 227–229.
Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2001a). Optical measurement of
microscopic torques. J. Mod. Opt. 48, 405–413.
Nieminen, T. A., Rubinsztein-Dunlop, H., Heckenberg, N. R., and Bishop, A. I. (2001b). Numerical
modelling of optical trapping. Comput. Phys. Commun. 142, 468–471.
Nieminen, T. A., Rubinsztein-Dunlop, H., and Heckenberg, N. R. (2003a). Calculation of the T-matrix:
General considerations and application of the point-matching method. J. Quant. Spectrosc. Radiat.
Transfer 79–80, 1019–1029.
Nieminen, T. A., Rubinsztein-Dunlop, H., and Heckenberg, N. R. (2003b). Multipole expansion of
strongly focussed laser beams. J. Quant. Spectrosc. Radiat. Transfer 79–80, 1005–1017.
Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. R. (2004a). Angular momentum of a
strongly focussed Gaussian beam. arXiv:physics/0408080.
Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. R. (2004b). Computational model-
ling of optical tweezers. Proc. SPIE 5514, 514–523.
Nieminen, T. A., Parkin, S. J., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2004c). Optical torque
and symmetry. Proc. SPIE 5514, 254–263.
Nieminen, T. A., Higuet, J., Knöner, G., Loke, V. L. Y., Parkin, S., Singer, W., Heckenberg, N. R., and
Rubinsztein-Dunlop, H. (2006). Optically driven micromachines: Progress and prospects. Proc. SPIE
6038, 237–245.
O’Neil, A. T., MacVicar, I., Allen, L., and Padgett, M. J. (2002). Intrinsic and extrinsic nature of the
orbital angular momentum of a light beam. Phys. Rev. Lett. 88, 053601.
Padgett, M. J. (2004). The mechanism for energy transfer in the rotational frequency shift of a light
beam. J. Opt. A: Pure Appl. Opt. 6, S263–S265.
Pfeifer, R. N. C., Nieminen, T. A., Heckenberg, N. R., and Rubinsztein-Dunlop, H. (2006). Two
controversies in classical electromagnetism. Proc. SPIE 6326, 93260H.
Phillips, W. D. (1998). Laser cooling and trapping of neutral atoms. Rev. Mod. Phys. 70, 721–741.
Poynting, J. H. (1884). On the transfer of energy in the electromagnetic field. Philos. Trans. R. Soc.
Lond. 175, 343–361.
Poynting, J. H. (1909). The wave motion of a revolving shaft, and a suggestion as to the angular
momentum in a beam of circularly polarised light. Proc. R. Soc. Lond. A 82, 560–567.
Quake, S. R., Babcock, H., and Chu, S. (1997). The dynamics of partially extended single molecules of
DNA. Nature 388, 151–154.
236 Timo A. Nieminen et al.

Ren, K. F., Gréhan, G., and Gouesbet, G. (1996). Prediction of reverse radiation pressure by
generalized Lorenz-Mie theory. Appl. Opt. 35, 2702–2710.
Rubinsztein-Dunlop, H., Nieminen, T. A., Friese, M. E. J., and Heckenberg, N. R. (1998). Optical
trapping of absorbing particles. Adv. Quantum Chem. 30, 469–492.
Sato, S., Ishigure, M., and Inaba, H. (1991). Optical trapping and rotational manipulation of micro-
scopic particles and biological cells using higher-order mode Nd:YAG laser beams. Electron. Lett. 27,
1831–1832.
Stewart, A. M. (2005). Angular momentum of the electromagnetic field: The plane wave paradox
resolved. Eur. J. Phys. 26, 635–641.
Stratton, J. A. (1941). ‘‘Electromagnetic Theory.’’ McGraw-Hill, New York.
Ukita, H., and Nagatomi, K. (1997). Theoretical demonstration of a newly designed micro-rotator
driven by optical pressure on a light incident surface. Opt. Rev. 4, 447–449.
van de Hulst, H. C. (1957). ‘‘Light Scattering by Small Particles.’’ Wiley, New York.
Visscher, K., and BrakenhoV, G. J. (1992). Theoretical study of optically induced forces on spherical
particles in a single beam trap. I. Rayleigh scatterers. Optik 89, 174–180.
Waterman, P. C. (1965). Matrix formulation of electromagnetic scattering. Proc. IEEE 53, 805–812.
Waterman, P. C. (1971). Symmetry, unitarity, and geometry in electromagnetic scattering. Phys.
Rev. D 3, 825–839.
Wright, W. H., Sonek, G. J., Tadir, Y., and Berns, M. W. (1990). Laser trapping in cell biology. IEEE
J. Quantum Electron. 26, 2148–2157.
Wright, W. H., Sonek, G. J., and Berns, M. W. (1993). Radiation trapping forces on microspheres with
optical tweezers. Appl. Phys. Lett. 63, 715–717.
Wright, W. H., Sonek, G. J., and Berns, M. W. (1994). Parametric study of the forces on microspheres
held by optical tweezers. Appl. Opt. 33, 1735–1748.
Zambrini, R., and Barnett, S. M. (2005). Local transfer of optical angular momentum to matter. J. Mod.
Opt. 52, 1045–1052.